How do we decide who should be released from prison, when, and under what circumstances? Arriving at a sound and reliable judgement on the basis of offender's past actions is difficult and the scientific literature argues that neither treatment nor supervision has demonstrable effects of recidivism. The authors of this book argue that community risk management can be improved by combining what is already known in three areas of inquiry: the prediction of violence; the study of clinical decision-making and clinical judgement; and the literature on treatment outcome and programme evaluation. On the basis of the literature and of their own research on mentally disordered offenders, sex offenders, foresetters, and psychopaths, the authors argue that an actuarial method of prediction is considerably superior to its alternatives.